The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script (scripts/aegis-workspace.py) to initialize the project workspace and manage the documentation index. This is an expected behavior for a developer-oriented automation tool and does not involve remote code downloading or unauthorized system changes.
[DATA_EXPOSURE]: No sensitive file access (such as credentials, SSH keys, or environment variables) was identified. The skill's file system operations are restricted to creating and managing implementation plans and documentation within the local docs/aegis/ directory.
[PROMPT_INJECTION]: The skill does not contain any patterns attempting to override agent behavior, bypass safety filters, or extract system instructions. It includes internal safeguards such as a 'Self-Review' step and a dedicated 'Plan Document Reviewer' subagent to ensure plan quality and alignment with specifications.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external data in the form of specifications and requirements to generate implementation plans. While this represents a standard injection surface, the skill mitigates risks through a structured planning hierarchy, explicit self-review checklists, and a human-in-the-loop choice for the execution handoff (Subagent-Driven vs. Inline).

writing-plans